Parallel-pipeline-based traversal unit for hardware-accelerated ray tracing

Jin Woo Kim, Won Jong Lee, Min Woo Lee, Tack-Don Han

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

In this work, we propose a novel parallel-pipeline traversal unit for hardware-based ray tracing, which can reduce latency and increase cache locality. Owing to the high memory bandwidth and computation requirements of ray-tracing operations such as traversal and intersection tests, recent studies have focused on the development of a hardware-based traversal and intersection-test unit[Nah et al. 2011][Lee et al. 2012]. Existing hardware engines are based on a single deep pipeline structure that increases the throughput of ray processing per unit time. However, traversal operations involve non-deterministic changes in the states of a ray. Therefore, in some cases, the ray may be unnecessarily transferred between pipeline stages, thereby increasing the overall latency. In order to solve this problem, we propose a parallel traversal unit having a pipeline per state. Our results show that the proposed system is up to 30% more efficient than a single-pipeline system because it decreases average latency per ray and increases cache efficiency. Copyright is held by the author / owner(s).

Original languageEnglish
Title of host publicationSIGGRAPH Asia 2012 Posters, SA 2012
DOIs
Publication statusPublished - 2012 Dec 31
EventSIGGRAPH Asia 2012 Posters, SA 2012 - Singapore, Singapore
Duration: 2012 Nov 282012 Dec 1

Publication series

NameSIGGRAPH Asia 2012 Posters, SA 2012

Other

OtherSIGGRAPH Asia 2012 Posters, SA 2012
CountrySingapore
CitySingapore
Period12/11/2812/12/1

Fingerprint

Ray tracing
Pipelines
Hardware
Throughput
Engines
Bandwidth
Data storage equipment
Processing

All Science Journal Classification (ASJC) codes

  • Computer Graphics and Computer-Aided Design
  • Computer Vision and Pattern Recognition

Cite this

Kim, J. W., Lee, W. J., Lee, M. W., & Han, T-D. (2012). Parallel-pipeline-based traversal unit for hardware-accelerated ray tracing. In SIGGRAPH Asia 2012 Posters, SA 2012 [42] (SIGGRAPH Asia 2012 Posters, SA 2012). https://doi.org/10.1145/2407156.2407203
Kim, Jin Woo ; Lee, Won Jong ; Lee, Min Woo ; Han, Tack-Don. / Parallel-pipeline-based traversal unit for hardware-accelerated ray tracing. SIGGRAPH Asia 2012 Posters, SA 2012. 2012. (SIGGRAPH Asia 2012 Posters, SA 2012).
@inproceedings{06972030b3cc4d4a896c5cd90e9509c3,
title = "Parallel-pipeline-based traversal unit for hardware-accelerated ray tracing",
abstract = "In this work, we propose a novel parallel-pipeline traversal unit for hardware-based ray tracing, which can reduce latency and increase cache locality. Owing to the high memory bandwidth and computation requirements of ray-tracing operations such as traversal and intersection tests, recent studies have focused on the development of a hardware-based traversal and intersection-test unit[Nah et al. 2011][Lee et al. 2012]. Existing hardware engines are based on a single deep pipeline structure that increases the throughput of ray processing per unit time. However, traversal operations involve non-deterministic changes in the states of a ray. Therefore, in some cases, the ray may be unnecessarily transferred between pipeline stages, thereby increasing the overall latency. In order to solve this problem, we propose a parallel traversal unit having a pipeline per state. Our results show that the proposed system is up to 30{\%} more efficient than a single-pipeline system because it decreases average latency per ray and increases cache efficiency. Copyright is held by the author / owner(s).",
author = "Kim, {Jin Woo} and Lee, {Won Jong} and Lee, {Min Woo} and Tack-Don Han",
year = "2012",
month = "12",
day = "31",
doi = "10.1145/2407156.2407203",
language = "English",
isbn = "9781450319119",
series = "SIGGRAPH Asia 2012 Posters, SA 2012",
booktitle = "SIGGRAPH Asia 2012 Posters, SA 2012",

}

Kim, JW, Lee, WJ, Lee, MW & Han, T-D 2012, Parallel-pipeline-based traversal unit for hardware-accelerated ray tracing. in SIGGRAPH Asia 2012 Posters, SA 2012., 42, SIGGRAPH Asia 2012 Posters, SA 2012, SIGGRAPH Asia 2012 Posters, SA 2012, Singapore, Singapore, 12/11/28. https://doi.org/10.1145/2407156.2407203

Parallel-pipeline-based traversal unit for hardware-accelerated ray tracing. / Kim, Jin Woo; Lee, Won Jong; Lee, Min Woo; Han, Tack-Don.

SIGGRAPH Asia 2012 Posters, SA 2012. 2012. 42 (SIGGRAPH Asia 2012 Posters, SA 2012).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Parallel-pipeline-based traversal unit for hardware-accelerated ray tracing

AU - Kim, Jin Woo

AU - Lee, Won Jong

AU - Lee, Min Woo

AU - Han, Tack-Don

PY - 2012/12/31

Y1 - 2012/12/31

N2 - In this work, we propose a novel parallel-pipeline traversal unit for hardware-based ray tracing, which can reduce latency and increase cache locality. Owing to the high memory bandwidth and computation requirements of ray-tracing operations such as traversal and intersection tests, recent studies have focused on the development of a hardware-based traversal and intersection-test unit[Nah et al. 2011][Lee et al. 2012]. Existing hardware engines are based on a single deep pipeline structure that increases the throughput of ray processing per unit time. However, traversal operations involve non-deterministic changes in the states of a ray. Therefore, in some cases, the ray may be unnecessarily transferred between pipeline stages, thereby increasing the overall latency. In order to solve this problem, we propose a parallel traversal unit having a pipeline per state. Our results show that the proposed system is up to 30% more efficient than a single-pipeline system because it decreases average latency per ray and increases cache efficiency. Copyright is held by the author / owner(s).

AB - In this work, we propose a novel parallel-pipeline traversal unit for hardware-based ray tracing, which can reduce latency and increase cache locality. Owing to the high memory bandwidth and computation requirements of ray-tracing operations such as traversal and intersection tests, recent studies have focused on the development of a hardware-based traversal and intersection-test unit[Nah et al. 2011][Lee et al. 2012]. Existing hardware engines are based on a single deep pipeline structure that increases the throughput of ray processing per unit time. However, traversal operations involve non-deterministic changes in the states of a ray. Therefore, in some cases, the ray may be unnecessarily transferred between pipeline stages, thereby increasing the overall latency. In order to solve this problem, we propose a parallel traversal unit having a pipeline per state. Our results show that the proposed system is up to 30% more efficient than a single-pipeline system because it decreases average latency per ray and increases cache efficiency. Copyright is held by the author / owner(s).

UR - http://www.scopus.com/inward/record.url?scp=84871544123&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84871544123&partnerID=8YFLogxK

U2 - 10.1145/2407156.2407203

DO - 10.1145/2407156.2407203

M3 - Conference contribution

SN - 9781450319119

T3 - SIGGRAPH Asia 2012 Posters, SA 2012

BT - SIGGRAPH Asia 2012 Posters, SA 2012

ER -

Kim JW, Lee WJ, Lee MW, Han T-D. Parallel-pipeline-based traversal unit for hardware-accelerated ray tracing. In SIGGRAPH Asia 2012 Posters, SA 2012. 2012. 42. (SIGGRAPH Asia 2012 Posters, SA 2012). https://doi.org/10.1145/2407156.2407203